Appliance use within a domestic property typically accounts for approximately 20% of the energy demand, however detailed understanding of how much energy is used by each individual appliance is often not known due the absence of granular energy monitoring. Without meter readings at an individual appliance level, analysis techniques such as non-intrusive load monitoring (NILM) can be used. NILM is the procedure that infers energy consumption of individual appliances given the aggregate energy consumption at household level. Various NILM techniques exist ranging from simpler models such as Combinatorial Optimization, through the Factorial Hidden Markov Model, to complex solutions such as Convolutional Neural Networks which require training using ground truth data . In this blog we show briefly how such data can be captured and how energy signatures can be recognised.
Controlling how much energy is used in a domestic dwelling offers many benefits including the potential for cost and carbon emission reductions, and greater security against future price increases or energy shortages. When it comes to reducing energy use in our homes an energy management approach can be used, typically involving several steps: monitoring, analysing, acting and tracking.
- Monitoring - taking meter readings and storing the data
- Analysing - the data to identify how much energy you’re consuming, when you’re consuming it, which appliances are consuming it, and opportunities to reduce energy use.
- Action – implementing energy savings retrofit opportunities to reduce routinely wasted and inefficient use of energy.
- Tracking - record energy use to monitor the impact of energy efficiency retrofit actions.
Appliance energy monitoring can be used to identify opportunities for demand reduction (e.g. by upgrading to more efficient appliances or through behaviour change). This data can also be used to quantify the suitability, impact and performance of solar PV coupled battery storage on a domestic scale and to begin a data-driven approach to the responsive smart home. A range of monitoring devices are available. We used an energy monitoring device to demonstrate how data analysis can be conducted in an average property.
The energy monitor can be easily installed onto the electricity supply of the dwelling. The data was accessed through the home router. A firewall and port forward rule was added to the router to enable a remote connection via SSH. It was configured with a dynamic DNS client to provide a persistent remote connection address. A python script runs on cron on a server at City Science’s office to process meter readings automatically over SSH, and pipe the data into a local database. This database allows City Science to organise, interrogate, analyse and manipulate large datasets quickly using a standardised procedure which can be scaled-up to multiple devices with ease. Once the data is extracted we can begin looking at appliance signatures.
Analysis of energy consumption
Figure 1 shows the energy consumption collected over the course of a week-day in February 2017. You can see from the figure how different appliances create different signatures that can be recognised through non-intrusive processes. Using this data we can begin to analyse and prioritise actions for this property.
Over the week, the total consumption was approximately 6kWh. From a purely qualitative point of view, we can start to build a sense of how energy is used throughout the property, for example:
- two occupants showered for 5 minutes each just after 6.30am.
- a dishwasher cycle was started at 7.30am and finished at 9.30am - the peaks observed are water heating at the beginning and end of the cycle.
- energy use above the baseload ceases after approximately 8.30am – it is likely that there are no occupants in the property at this point. The house is unoccupied during the day, baseload energy use is observed at approximately 35W.
- no significant electrical demand is observed during the night, indicating the properly is likely heated using gas rather than electrically with night storage heaters.
- electricity is periodically demanded by the boiler to power the central heating pump when the internal temperature of the house reaches the set-back heating temperature.
While there are high power appliances in the property such as the electric shower, there is very little wastage since the baseload requirement is low and very little energy is used outside of occupied hours. The data from the installation uncovered two inefficient appliances that could be upgraded to provide energy savings – two T12 fluorescent tube lights in the garage, used for an hour a day, which could be upgraded to LED, and a 5-year-old plasma television which could also be upgraded to LED.
Towards the smart home
The data can also be used to assess the potential smart energy benefits. It is clear from the timing of energy use that installing a stand-alone solar PV system to this property wouldn’t be beneficial from the point of view of maximising self-consumption of the generated electricity (Figure 2). Very little electricity is consumed at the time of PV generation and no loads exist that can be shifted into the time of generation such as hot water heating on a smart switch. However, a battery-coupled solar PV system could work here to increase the self-utilisation of solar PV energy by storing energy during the day, instead of exporting to the grid, for use in the evening.
Energy monitoring is powerful tool for identifying areas where appliances may be inefficient, opportunities to use energy more efficiently and for evaluating the impact of new technologies. Similar energy monitoring and analysis techniques can be applied to gas consumption, or across commercial buildings and industrial processes. Cheap, flexible and non-intrusive options exist that can enable the collection and analysis of this data today. City Science’s research and data analysis capability is technology agnostic and our approach to problems elicits maximum insight from data. To understand how energy monitoring techniques might help create value for your property or business, please get in touch.
Anthony Vickers, MEng, is our Energy Consultant, he helped successfully deliver a commercial off-grid wind farm in Patagonia and co-author of the Exeter Energy Independence Report